Measure and Manage

AI Agent Analytics 

AI Agent Analytics provides a comprehensive overview of the AI Agent’s performance, helping users assess its effectiveness in handling customer interactions. This dashboard provides valuable insights into how the AI Agent is performing, highlighting areas where it excels and identifying opportunities for improvement.

Analyzing key metrics offers valuable insights, allowing users to identify patterns that optimize the AI Agent’s performance and improve the overall customer experience.

The AI Agent Analytics displays the following metrics and details over the course of the previous month, the previous week, or the current week:

  • All conversations

  • Resolved conversations

  • Abandoned conversations

  • Active Conversations

AI Agent Analytics

AI Agent Conversations

The AI Agent Conversations section is visible only after saving the AI Agent. This section enables users to review past interactions, providing valuable insights into how the AI Agent engages with customers. By analyzing conversation details, users can identify successful interactions that can be replicated, as well as areas where improvements are needed.

The Conversations section is divided into three main areas that help users review, analyze, and manage chat interactions.

1. Conversation List

The left panel displays a list of all conversations. Users can select any conversation to view its details.

  • Search Messages: The search option allows users to quickly locate conversations using the user name, phrases from the conversation, or session ID.
  • Filters: Users can refine the conversation list using the following filters:

    • Status: Resolved, Abandoned, Active, or All

    • Topics: Filter by topics the AI Agent is trained on

    • Feedback: All, Positive, or Negative

    • Portal: Includes search and pagination, available when integrated with eGain Knowledge Base

    • Certified: All, Certified, or No Certified

2. Conversation Transcript

The middle panel displays the selected conversation transcript along with the session ID and the state of the conversation (Active, Resolved, Abandoned, or Escalated).

  • Users can export individual transcripts or use Export Transcripts to download multiple conversations as a zip file.

  • Exported transcripts include chat messages along with metadata such as session IDs, message timestamps (in the exporting user’s timezone), sender details with personalization IDs, source titles and IDs, certified answer titles and IDs, feedback scores, and session attributes such as portal name, username, conversation start time, and location. For bulk exports, the system generates a separate Session Attributes file along with the Chat Transcripts file. Bulk exports also respect any filters applied at the time of export.

3. Conversation Details

The right panel provides detailed information about the selected conversation, enabling deeper analysis and review.

  • User: Displays the type of participant (for example, visitor).

  • Conversation started at: Shows the date and time when the conversation began.

  • Location: Indicates the geographic location of the user.

  • Sources AI Agent Referred To: Lists the knowledge sources or content used by the AI Agent while responding during the conversation.

  • Feedback: Displays user feedback collected during the session, including counts of helpful and unhelpful responses.

  • Average AI Agent Response Time: Shows the average time taken by the AI Agent to respond during the conversation, helping assess response performance.

  • Questions in the Conversation: Lists the questions asked by the user throughout the interaction, providing insight into user intent and behavior.

AI Agent Conversations